Tearsheet

Category Value Note
Metric
Period frequency Meta 1D Sampling frequency used for annualization. Smaller periods are generally more granular (but can be noisier).
Benchmark Asset Meta BTC Column name of the benchmark asset used for alpha/beta and benchmark charts (if provided).
Fee % Meta 0.00025 Trading fee rate applied to order notional (decimal units; e.g. 0.001 = 10 bps).
Slippage % Meta 0.001 Slippage applied against the trader on execution prices (decimal units; e.g. 0.001 = 10 bps).
Init Cash Meta 10000.0 Initial cash (starting equity) used for the simulation.
Trading Days Year Meta 365 Trading days per year used for annualization.
Risk Free Rate Meta 0.03 Annual risk-free rate used for Sharpe/Sortino (decimal units).
Simulation start date Meta 2023-08-09 00:00:00+00:00 First timestamp in the simulation index. Earlier start dates generally make estimates more statistically stable.
Simulation end date Meta 2025-12-06 00:00:00+00:00 Last timestamp in the simulation index. More recent end dates generally better reflect current market conditions.
First transaction date Meta 2023-08-09 00:00:00+00:00 First timestamp with any executed trade. Earlier is generally better (less time inactive), depending on the strategy.
Annualized return % Performance 0.954832 Geometric mean return annualized (decimal units). Higher is generally better, but interpret alongside risk and drawdowns.
Annualized volatility Performance 0.486667 Sample standard deviation of returns annualized (decimal units). Lower is generally better for a given return level. Uses Bessel's correction (ddof=1) per industry standard.
Annualized Sharpe Performance 1.900338 Annualized excess return divided by annualized volatility (sample statistics). Higher is generally better (rule of thumb: >1 is good, >2 is strong).
Max drawdown (equity) % Performance -30.015971 Worst peak-to-trough % decline in equity. Less negative (closer to 0) is generally better.
Max drawdown (PnL) % Performance -213.575379 Worst drawdown of cumulative PnL relative to prior PnL peak. Less negative (closer to 0) is generally better.
Total return % Performance 373.578518 Ending equity / initial cash minus 1, expressed in percent. Higher is generally better.
Funding earnings Costs 10.013426 Sum of funding payments (positive means net earned). Higher is generally better; negative values mean funding cost.
Fees Costs 1310.870193 Sum of trading fees paid. Lower is generally better.
Annual turnover Costs 52.394012 Average per-period one-sided turnover annualized (not percent), computed as min(total buys, total sells) / equity before trading. Lower is generally better (less trading/costs), unless the strategy requires frequent rebalancing.
Total order count Costs 52421 Count of non-zero notional orders executed. Lower generally means less trading (and costs), but too low can indicate inactivity.
Average order notional Costs 100.02634 Mean absolute notional per executed order. Good depends on liquidity and constraints; too large can be hard to execute.
Gross exposure mean % Exposure 128.741476 Average sum(|positions|) as % of equity. Lower generally means less leverage; values above 100% indicate leveraged exposure.
Gross exposure median % Exposure 125.958446 Median sum(|positions|) as % of equity. Lower generally means less leverage; values above 100% indicate leveraged exposure.
Gross exposure max % Exposure 400.43439 Maximum sum(|positions|) as % of equity. Lower generally means tighter leverage control; very high peaks imply occasional high leverage.
Net exposure mean % Exposure -0.848327 Average signed exposure as % of equity. Closer to 0 is generally more market-neutral; positive means net long, negative net short.
Net exposure median % Exposure -0.132525 Median signed exposure as % of equity. Closer to 0 is generally more market-neutral; positive means net long, negative net short.
Net exposure max % Exposure 128.13548 Max absolute signed exposure as % of equity. Lower absolute values generally mean better exposure control.
Alpha Benchmark 0.787718 Annualized intercept vs benchmark excess returns (CAPM-style, sample statistics). Higher is generally better; near 0 implies little outperformance after adjusting for beta.
Beta Benchmark -0.057862 Slope vs benchmark excess returns (CAPM-style, sample covariance/variance). Values near 1 behave like the benchmark; values near 0 have low benchmark sensitivity.
Benchmark annualized return % Benchmark 60.570186 Benchmark geometric mean return annualized (percent units). Higher is generally better, but depends on your benchmark choice and sample.
Active annual return % Benchmark 20.026842 Arithmetic mean of (strategy - benchmark) period returns annualized (percent units). Uses arithmetic (not geometric) mean to match tracking error calculation. Higher is generally better; negative means underperformance vs the benchmark.
Tracking error Benchmark 0.697594 Sample std dev of active returns annualized (decimal units). Lower means closer to the benchmark; higher means more active risk.
Information ratio Benchmark 0.287084 Active annual return divided by tracking error. Higher is generally better (rule of thumb: >0.5 is decent, >1 is strong).
R2 vs benchmark Benchmark 0.003165 Squared correlation of returns vs benchmark returns. Higher means the benchmark explains more of the returns; lower implies more idiosyncratic behavior.
Calmar ratio Distribution 3.18108 Annualized return divided by absolute max equity drawdown. Higher is generally better (more return per unit of drawdown).
Skewness Distribution 2.387562 Skewness of period returns distribution. More positive skewness is often preferred (more upside tail), all else equal.
Kurtosis Distribution 27.531241 Excess kurtosis of period returns distribution (normal distribution = 0). Higher values indicate fatter tails; lower (negative) values indicate thinner tails.
Best period return Distribution 0.286998 Maximum single-period return. Higher is generally better, but interpret alongside worst-period and drawdowns.
Worst period return Distribution -0.127536 Minimum single-period return. Less negative (closer to 0) is generally better.
Hit rate Distribution 0.527059 Fraction of non-zero return periods that are positive. Higher is generally better.
Avg win Distribution 0.018113 Mean return of positive-return periods. Higher is generally better.
Avg loss Distribution -0.015632 Mean return of negative-return periods. Less negative (closer to 0) is generally better.
Profit factor Distribution 1.291312 Sum of wins divided by absolute sum of losses. Higher is generally better; values >1 mean wins outweigh losses.
Max drawdown duration (periods) Distribution 130 Longest consecutive underwater duration in periods. Shorter is generally better (capital recovers faster).
Time to recovery (periods) Distribution 131 Periods from drawdown peak to recovering the prior peak. Shorter is generally better.
Average holding period Portfolio 26.710457 Average consecutive periods with a non-zero position per asset. Good depends on the strategy; shorter implies more trading, longer implies lower turnover.
Costs % gross pnl Portfolio 14.926118 Fees+slippage as % of gross PnL (before costs). Lower is generally better; near 0 means costs are small relative to edge.
Funding % total pnl Portfolio 0.026804 Funding as % of net PnL. Lower absolute values are generally better; large magnitudes mean funding dominates PnL.
Average funding settled Portfolio 0.011767 Average funding payment per period. Positive is generally better; negative means funding paid on average.
Max abs weight Portfolio 2.020014 Maximum absolute target weight across assets/periods. Lower is generally better (less concentration/leverage), given the strategy's intent.
Mean abs weight Portfolio 0.040634 Mean absolute target weight across assets/periods. Lower is generally better (less aggregate risk), given the strategy's intent.
Annualized Sortino Risk 3.178483 Annualized excess return divided by annualized downside deviation. Higher is generally better; focuses on downside risk unlike Sharpe which penalizes upside volatility.
Downside deviation Risk 0.290967 Sample std dev of negative returns annualized (decimal units). Lower is generally better; measures downside risk only.
VaR 95% Risk -0.033384 Value at Risk at 95% confidence level (5th percentile of returns). Less negative (closer to 0) is generally better; worst expected loss in 19 out of 20 periods.
CVaR 95% Risk -0.048141 Conditional Value at Risk at 95% confidence level (mean of returns below VaR). Less negative (closer to 0) is generally better; average loss when VaR is exceeded.
Omega Ratio Risk 1.291312 Probability-weighted ratio of gains above threshold vs losses below threshold (uses 0 as threshold). Higher is generally better; values >1 mean gains outweigh losses.
Gain-to-Pain Ratio Risk 0.127138 Sum of returns divided by sum of absolute returns. Higher is generally better; measures return per unit of total volatility.
Ulcer Index Risk 5.993044 RMS (root mean square) of drawdowns, annualized. Lower is generally better; alternative drawdown-based risk measure that penalizes depth and duration.
Weight IC mean (next) Signal 0.022116 Time-average cross-sectional correlation between weights at t and next-period asset returns (close-to-close), computed over the active universe (non-zero weights) each period.
Weight IC t-stat (next) Signal 1.931289 t-stat of the time series of per-period weight IC values. Higher absolute values suggest more statistically reliable alignment (not a guarantee).
Weight Rank IC mean (next) Signal -0.013531 Time-average Spearman-style (rank) IC between weights and next-period asset returns, computed over the active universe (non-zero weights) each period.
Weight Rank IC t-stat (next) Signal -1.191637 t-stat of the time series of per-period weight rank IC values.
Top-bottom decile spread mean (next) Signal 0.005087 Time-average next-period return spread between the top and bottom weight deciles within the active universe (assets with non-zero weights) each period.
Top-bottom decile spread t-stat (next) Signal 2.234272 t-stat of the time series of top-minus-bottom decile spreads.
Weighted long hit rate mean (next) Signal 0.476172 Average fraction of long gross weight placed in assets that have positive next-period returns (weights within each period). Higher is generally better.
Weighted short hit rate mean (next) Signal 0.515359 Average fraction of short gross weight placed in assets that have negative next-period returns (weights within each period). Higher is generally better.
Forward return per gross mean (next) Signal 0.001856 Average of (Σ w_t,i r_{t+1,i}) / (Σ |w_t,i|) each period. Normalizes for varying leverage and compares return per unit of gross weight.
Forward return selection per gross mean (next) Signal 0.001625 Average of the cross-sectional selection component of Σ w_t,i r_{t+1,i}, normalized by gross weight (active universe, next-period). Higher is generally better.
Forward return selection per gross t-stat (next) Signal 2.675715 t-stat of the time series of per-period selection-per-gross values.
Forward return directional per gross mean (next) Signal 0.000231 Average of the directional component (net weight × mean next return of the active universe), normalized by gross weight (next-period). Magnitude near 0 indicates little directional dependence.
Forward return directional per gross t-stat (next) Signal 1.328641 t-stat of the time series of per-period directional-per-gross values.
Gross weight mean Signal 1.280167 Average gross weight (Σ |w_t,i|) across periods with available next returns. Higher implies more leverage/total exposure in the signal.
Directionality mean Signal 0.004842 Average net-to-gross ratio (Σ w_t,i) / (Σ |w_t,i|). Values near 0 indicate market-neutral; positive is net long; negative net short.

Equity

Drawdown

Returns Distribution

Signal